# Constrained Decoding for Computationally Efficient Named Entity Recognition Taggers

arXiv: https://arxiv.org/abs/2010.04362 · HTML: https://mlsanity.com/paper/2010.04362/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Oct 2020.

## Built on by (most cited first, 8 of 8)

- [KPI-BERT: A Joint Named Entity Recognition and Relation Extraction Model for Financial Reports](https://mlsanity.com/paper/2208.02140/) (2022)
- [KPI-EDGAR: A Novel Dataset and Accompanying Metric for Relation Extraction from Financial Documents](https://mlsanity.com/paper/2210.09163/) (2022)
- [SeqScore: Addressing Barriers to Reproducible Named Entity Recognition Evaluation](https://mlsanity.com/paper/2107.14154/) (2021)
- [Intent Features for Rich Natural Language Understanding](https://mlsanity.com/paper/2104.08701/) (2021)
- [Optimizing small BERTs trained for German NER](https://mlsanity.com/paper/2104.11559/) (2021)
- [Improving Text-to-SQL Semantic Parsing with Fine-grained Query Understanding](https://mlsanity.com/paper/2209.14415/) (2022)
- [PathCRF: Ball-Free Soccer Event Detection via Possession Path Inference from Player Trajectories](https://mlsanity.com/paper/2602.12080/) (2026)
- [Meddies-PII: A Multilingual Framework for Personally Identifiable Information Extraction in Clinical De-identification](https://mlsanity.com/paper/2609.12544/) (2026)

## Abstract (arXiv)

Current state-of-the-art models for named entity recognition (NER) are neural models with a conditional random field (CRF) as the final layer. Entities are represented as per-token labels with a special structure in order to decode them into spans. Current work eschews prior knowledge of how the span encoding scheme works and relies on the CRF learning which transitions are illegal and which are not to facilitate global coherence. We find that by constraining the output to suppress illegal transitions we can train a tagger with a cross-entropy loss twice as fast as a CRF with differences in F1 that are statistically insignificant, effectively eliminating the need for a CRF. We analyze the dynamics of tag co-occurrence to explain when these constraints are most effective and provide open source implementations of our tagger in both PyTorch and TensorFlow.